Update Tesseract (to 4.0!), PDFBox

overview · Jan 25, 2019 · 0268f60 · 0268f60
1 parent 42fee7d
commit 0268f60
Show file tree

Hide file tree

Showing 5 changed files with 13 additions and 12 deletions.
diff --git a/build.sbt b/build.sbt
@@ -2,7 +2,7 @@
 
 name := "pdfocr"
 
-version := "0.0.10"
+version := "0.0.11"
 
 homepage := Some(url("https://github.com/overview/pdfocr"))
 
@@ -21,18 +21,18 @@ fork in (Compile, run) := true // Main calls System.exit() to give proper return
 
 // Compile settings
 
-scalaVersion := "2.12.6"
+scalaVersion := "2.12.8"
 
 scalacOptions += "-deprecation"
 
 libraryDependencies ++= Seq(
-  "org.bouncycastle" % "bcmail-jdk15on" % "1.59", // https://pdfbox.apache.org/2.0/dependencies.html
-  "org.bouncycastle" % "bcprov-jdk15on" % "1.59", // https://pdfbox.apache.org/2.0/dependencies.html
-  "org.bouncycastle" % "bcpkix-jdk15on" % "1.59", // https://pdfbox.apache.org/2.0/dependencies.html
+  "org.bouncycastle" % "bcmail-jdk15on" % "1.60", // https://pdfbox.apache.org/2.0/dependencies.html
+  "org.bouncycastle" % "bcprov-jdk15on" % "1.60", // https://pdfbox.apache.org/2.0/dependencies.html
+  "org.bouncycastle" % "bcpkix-jdk15on" % "1.60", // https://pdfbox.apache.org/2.0/dependencies.html
   "com.github.jai-imageio" % "jai-imageio-core" % "1.4.0", // for TIFF support
   "com.github.jai-imageio" % "jai-imageio-jpeg2000" % "1.3.0", // for JPEG2000 support
   "org.apache.pdfbox" % "jbig2-imageio" % "3.0.0",
-  "org.apache.pdfbox" % "pdfbox" % "2.0.9",
+  "org.apache.pdfbox" % "pdfbox" % "2.0.13",
   "org.scalatest" %% "scalatest" % "3.0.5" % "test",
   "org.mockito" % "mockito-core" % "2.18.3" % "test",
   "org.slf4j" % "jcl-over-slf4j" % "1.7.25" % "test", // So we can mute warnings during testing

diff --git a/src/main/scala/org/overviewproject/pdfocr/ocr/Tesseract.scala b/src/main/scala/org/overviewproject/pdfocr/ocr/Tesseract.scala
@@ -93,9 +93,10 @@ class Tesseract(val options: TesseractOptions) {
       try {
         val processBuilder = new ProcessBuilder(
           options.tesseractPath,
-          "-", "-",
           "-l", languages.map(_.getISO3Language).mkString("+"), // Languages
-          "-psm", "1",                                          // Page segmentation + orientation/script detection
+          "--psm", "1",                                         // Page segmentation + orientation/script detection
+          "--oem", "1",                                         // LTSM engine (Tesseract 4.0)
+          "-", "-",                                             // stdin, stdout
           "hocr"
         )
         processBuilder.start

diff --git a/...ources/expected-images/pdfPage-toImage-should-return-the-correct-image-data.png b/...ources/expected-images/pdfPage-toImage-should-return-the-correct-image-data.png
diff --git a/src/test/resources/fake-tesseract b/src/test/resources/fake-tesseract
@@ -4,14 +4,14 @@
 #
 # Spits out its invocation to stderr, then spits out its input to stdout.
 #
-# Special language values ($4) cause special results:
+# Special language values ($2) cause special results:
 #
 # * "zxx": retval 1, missing-language warning to stderr
 # * "osd": retval 0, missing-language warning to stderr
 # * "und": retval 1, other error message
 
 
-case "$4" in
+case "$2" in
   osd)
     # Put dummy data above and below the language message we grep for
     echo "blah" >&2

diff --git a/src/test/scala/org/overviewproject/pdfocr/ocr/TesseractSpec.scala b/src/test/scala/org/overviewproject/pdfocr/ocr/TesseractSpec.scala
@@ -17,12 +17,12 @@ class TesseractSpec extends UnitSpec {
 
   it("shells to Tesseract and collects stderr") {
     val result = tesseract.ocr(image, Seq(new Locale("en"))).futureValue
-    new String(result.standardError, "utf-8") must equal("- - -l eng -psm 1 hocr\n")
+    new String(result.standardError, "utf-8") must equal("-l eng --psm 1 --oem 1 - - hocr\n")
   }
 
   it("concatenates languages using +") {
     val result = tesseract.ocr(image, Seq(new Locale("en"), new Locale("fr"))).futureValue
-    new String(result.standardError, "utf-8") must equal("- - -l eng+fra -psm 1 hocr\n")
+    new String(result.standardError, "utf-8") must equal("-l eng+fra --psm 1 --oem 1 - - hocr\n")
   }
 
   it("sends Tesseract the image as BMP") {